On 02/23/2017 03:45 PM, Markus Armbruster wrote: > Signed-off-by: Markus Armbruster <arm...@redhat.com> > --- > qapi/qmp-dispatch.c | 14 ++++++++------ > 1 file changed, 8 insertions(+), 6 deletions(-) >
> @@ -41,15 +41,17 @@ static QDict *qmp_dispatch_check_obj(const QObject
> *request, Error **errp)
>
> if (!strcmp(arg_name, "execute")) {
> if (qobject_type(arg_obj) != QTYPE_QSTRING) {
> - error_setg(errp, "QMP input object member '%s' expects '%s'",
> - "execute", "string");
> + error_setg(errp,
> + "QMP input object member '%s' must be %s",
> + "execute", "a string");
Any reason we can't inline this to:
"QMP input object member 'execute' must be a string"
? It's not like we're translating the message.
> return NULL;
> }
> has_exec_key = true;
> } else if (!strcmp(arg_name, "arguments")) {
> if (qobject_type(arg_obj) != QTYPE_QDICT) {
> - error_setg(errp, "QMP input object member '%s' expects '%s'",
> - "arguments", "object");
> + error_setg(errp,
> + "QMP input object member '%s' must be %s",
> + "arguments", "an object");
and again
Then again, if you use my idea of a QAPI-generated visitor of each input
wire object, you'd get whatever error message qobject-input normally
gives, which may render these changes irrelevant.
At any rate, the new wordings are nicer, whether or not you inline
constant text.
Reviewed-by: Eric Blake <ebl...@redhat.com>
--
Eric Blake eblake redhat com +1-919-301-3266
Libvirt virtualization library http://libvirt.org
signature.asc
Description: OpenPGP digital signature
